MCP server with tools for number addition, letter counting, and Azure CLI command execution, plus prompts for code explanation and debugging
Server implements 3 tools with mixed quality. Tool naming follows verb_noun conventions (add_list_of_numbers, count_letter_in_text, run_azure_cli_command), which is positive. All tools have descriptions and input schemas with type definitions visible in the source code. However, descriptions are inconsistent in depth, parameter descriptions are minimal, and output schemas are not formally documented. The destructive Azure CLI tool has good safety guardrails (allowlist, character validation) but descriptions could better explain preconditions and error recovery paths. Tool annotations are present (readOnlyHint, destructiveHint, idempotentHint) which is a strength. Average description length across tools is ~110 characters, below the 194-char production baseline. Parameter descriptions are tersely generic ('List of integers to add', 'The text to analyze', 'The letter to count', 'The Azure CLI command'), they explain WHAT but not WHY or HOW to construct valid input.
Add a list of numbers and return structured result with metadata
Count occurrences of a letter in text with detailed analysis
Run an Azure CLI command with structured output and error handling
Output schemas are not formally documented. Tools return structured dicts but LLMs must infer the response shape from examples. Per pattern:tool, 100% of A+ tools have documented return types.
Parameter descriptions are minimal and generic. 'List of integers to add' does not explain constraints (min/max length, element bounds), use cases, or dependencies.
run_azure_cli_command tool description does not explicitly state that it is destructive, requires Azure credentials, or has side effects. Per pattern:command-tool, destructive tools MUST declare impact in the description text itself, not just via annotations. Description states 'Run an Azure CLI command' but does not warn about potential resource creation/deletion.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 56 | <=2025-11-25 | v2 |
| 2026-03-09 | F | 43 | - | v1 |
run_azure_cli_command accepts arbitrary command strings and filters via allowlist. While ALLOWED_COMMANDS restricts base commands, the error message 'Command contains potentially dangerous characters' is not actionable for LLM recovery. Per pattern:recovery-guide, errors must guide next steps. Should suggest: 'Use only base commands: version, account, group, resource, storage, vm, webapp, sql.'
Tool descriptions do not include prerequisite or dependency hints. For run_azure_cli_command, no mention of: 'Requires Azure CLI installed and user logged in via 'az login'. For count_letter_in_text, no note on case sensitivity. Per pattern:tool-description, discovery tools and stateful operations must document preconditions.